1. Field of the Invention
The present invention relates to data transmission systems and, more particularly, to multicarrier data transmission systems.
2. Description of the Related Art
A conventional voice-band modem can connect computer users end-to-end through the Public Switched Telephone Network (PSTN). However, the transmission throughput of a voice-band modem is limited to below 40 Kbps due to the 3.5 KHz bandwidth enforced by bandpass filters and codes at the PSTN interface points. On the other hand, the twisted-pair telephone subscriber loop of a computer user has a much wider usable bandwidth. Transmission systems based on local subscriber loops are generally called Digital Subscriber Lines (DSL).
One DSL technique for high-speed data communications is Asymmetrical Digital Subscriber Line (ADSL) signaling for the telephone loop which has been defined by standards as a communication system specification that provides a low-rate data stream from a residence to a telephone company's central office (upstream), and a high-rate data stream from the central office to the residence (downstream). The ADSL standard provides for operation without affecting conventional voice telephone communications, e.g., Plain Old Telephone Service (POTS). The ADSL upstream channel only provides simple control functions or low-rate data transfers. The high-rate downstream channel provides a much higher throughput. This asymmetrical information flow is desirable for applications such as video-on-demand (VOD).
An ADSL modem operates in a frequency range that is higher than the voice-band; this permits higher data rates. However, the twisted-pair subscriber line has distortion and losses which increase with frequency and line length. Thus, the maximum ADSL data rate is determined by the length of subscriber lines. The ADSL standard uses Discrete Multi-Tone (DMT) modulation with the DMT spectrum divided into two-hundred fifty-six 4.3125 kHz carrier bands and a Quadrature Amplitude Modulation (QAM) type of constellation is used to load a variable number of bits onto each carrier band independently of the other carrier bands.
Besides ADSL and its families (ADSL2, ADSL2+, all based on DMT modulations), another set of DSL technique for high-speed data communications over twisted-pair phone lines is known as Very High Speed Digital Subscriber Lines (VDSL) and VDSL2. VDSL and VDSL2 are intended to facilitate transmission rates greater than that offered by the ADSL families. The transmission schemes used with VDSL can be Discrete Multi-Tone (DMT) modulation or QAM modulation, whereas for VDSL2, its once again DMT modulation.
Digital communication systems transmit data from a transmitter over a channel to a receiver. In order to support reliable, high performance operation, digital communication systems often need to estimate the impulse response of the channel. The channel represents a communication medium from the transmitter to a far-end receiver or from the transmitter to a near-end receiver. The digital communication system can utilize the estimated channel impulse response for far-end or near-end noise cancellation schemes and far-end channel equalization.
Prior approaches to estimating a channel impulse response have been implemented in either the time domain or the frequency domain. In the case of time-domain channel estimation techniques, the estimated channel is convolved with the transmitted signal in an adaptive manner. However, such a solution produces only a single error signal that is used to update all taps of a finite impulse response filter that provides the estimated channel. This approach is complex and slow to converge.
Frequency-domain channel identification approaches are more common. One approach requires transforming time-domain signals to frequency-domain tones, training frequency-domain taps for a FIR filter that provides the estimated channel, and then finally converting the frequency-domain taps back to the time-domain channel estimate. This approach allows each tap to be independently trained and adapted in the frequency domain. However, the disadvantages of this frequency-domain approach are that additional hardware for fast Fourier transforms and inverse fast Fourier transforms are required on the receiver side, and that the training signals utilized must span the entire frequency bandwidth. Unfortunately, in some implementations of digital communication systems, there are restrictions on usage of certain frequencies for the purpose of training and thus the entire frequency bandwidth is sometimes not permitted to be used.
Another approach to estimating the channel response in the frequency domain can utilize a frequency-domain adaptive comb filter. In K. Van Acker, M. Moonen, T. Pollet, “Per-Tone Echo Cancellation for DMT-based system,” IEEE Transactions on Communications, Vol. 51, No. 9 (September 2003), a per tone echo cancellation structure enables the transformation of time-domain taps to the frequency-domain adaptive comb filter taps. See also Katleen Van Acker et al., “Per Tone Equalization for DMT-Based Systems,” IEEE Transactions on Communications, Vol. 49, No. 1 (January 2001). The frequency-domain adaptive comb filter taps can be directly trained to estimate the desired taps. The update of the adaptive comb filter taps for each of the frequency tones is based on an error signal.
FIG. 1 is a block diagram of a conventional per-tone channel equalizer 100. Although use of a time domain equalizer (TEQ) is common, in this embodiment, the channel equalization is performed in the frequency domain. The per-tone channel equalizer 100 includes a fast Fourier transform (FFT) 102, a scaler 104, a delay circuit 106, a subtractor 108, an adaptive comb filter 110, an adder 112, and a coefficient updating algorithm 114. Received signals, yn, are passed through the FFT 102 and supplied to the delay circuit 106. The subtractor 108 forms difference signals from the received signals and the delayed received signals. The difference signals from the subtractor 108 are fed to the adaptive comb filter 110. The coefficient updating algorithm 114 updates coefficients for taps of the adaptive comb filter 110 in accordance with a decision error signal. The output of the FFT 102 is adaptively scaled by the scaler 104 which is controlled by the coefficient updating algorithm 114. The adder 112 adds the scaled signals with the output of the adaptive comb filter 110. The output of the adder 112 is the equalizer output, Zi, and is calculated asZi=Yi·v0(i)+ACF(i)  Equation 1where ACF(i) is the adaptive comb filter output for the ith bin and is given by
                              ACF          ⁡                      (            i            )                          =                              ∑                          t              =              1                                      T              -              1                                ⁢                                    Δ              t                        ·                                          v                t                            ⁡                              (                i                )                                                                        Equation        ⁢                                  ⁢        2            where T is the number of taps, and the difference signals are given by the following equation.Δi=y-i−yN-i  Equation 3
In adapting the coefficients, a sequence of known training symbols is transmitted to a receiver and a Least Mean Square (LMS) based algorithm is applied to train up the coefficients. The detected error signal, ed(i), is computed ased(i)=Sd(i)−Zi  Equation 4where Sd(i) is the known transmitted training symbol.
The update of the tap coefficient v0(i) and the coefficients for the adaptive comb filter taps can be performed asv0(i)(k+1)=v0(i)(k)+μi·ed(i)·Yi*(k)  Equation 5vt(i)(k+1)=vt(i)(k)+μa·Δt·ed(i)for t=1, . . . , (T−1)  Equation 6where vt(i)(k) is the tth coefficient for the adaptive comb filter taps applied to the ith received tone signal; Δt is the difference signal; μa is the adaptation constant; ed(i) is the error signal for the ith tone, computed by Equation 4; finally, k represents the symbol index.
For a multicarrier system with multiple-inputs and multiple outputs, techniques for identifying cross channel coefficients due to far-end crosstalk (FEXT) are described, for example, in Starr, Sobara, Cioffi & Silverman, “DSL Advances,” Pearson Education, Inc., 2003. FIG. 2 is a block diagram of a FEXT cancellation system 200. The FEXT cancellation system 200 includes a FEXT channel estimator 202 that examines a received training sequence to adaptively estimate FEXT channel coefficients. Following training, the FEXT channel coefficients are supplied to a FEXT canceller 204 that cancels FEXT from incoming signals in accordance with the estimated FEXT channel coefficients.
Moreover, near-end crosstalk (NEXT) cancellation can also be conventionally provided. FIG. 3 is a block diagram of a NEXT cancellation system 300. The NEXT cancellation system 300 includes an NEXT channel estimator 302 and a subtractor 304. The NEXT channel estimator 302 adaptively estimates a NEXT channel. The subtractor 304 then subtracts the estimated NEXT noise from incoming signals. The NEXT cancellation process is summarized as
                                          E            NEXT                    ⁡                      (                          m              ,              i                        )                          =                              Y            i            m                    -                                    ∑                                                n                  =                  1                                                  n                  ≠                  m                                            L                        ⁢                                                            G                  i                                ⁡                                  (                                      m                    ,                    n                                    )                                            ·                                                X                  ne                                ⁡                                  (                                      n                    ,                    i                                    )                                                                                        Equation        ⁢                                  ⁢        7            where ENEXT(m,i) is the estimated near-end crosstalk estimation error for the mth user at the ith tone, if the far-end channel is silence; Yim is the received signal for the mth user at the ith tone; Gi(m,n) is the near-end crosstalk coefficient that couples the nth channel to the mth user channel at the ith tone; Xne(n,i) is the reference near-end transmitted signal of the nth channel at the ith tone.
The NEXT coefficients is updated asGi(m,n)(t+1)=Gi(m,n)(t)+a·ENEXT(m,i)·X*ne(n,i)  Equation 8where a is a small positive adaptation constant of value less than 1.
Conventionally, the training of the equalizer and the training of the FEXT estimator are performed at different times in the training protocol. If the equalizer is trained before the FEXT estimation gets started, the equalizer settings are obtained with the presence of the FEXT noise. The disadvantage in this case is that a higher noise level will slow down the equalizer training and increase the chance for mis-adjustment of the adaptive comb filter taps. On the other hand, if the FEXT estimator is trained before the equalizer, the presence of the inter-carrier interference and the inter-symbol interference disadvantageously deteriorates the FEXT channel estimation.
Thus, there remains a need for improved approaches to performing channel equalization as well as canceling interference in a digital communication system.